The Junior Worker Dilemma

I’m pretty darn good at getting what I want out of AI.

Why is that?

A good portion of it is because I’ve spent the last 3 1/2 years in deep with it. At first, like an 8-year-old with a new toy, I dove in. Then, once I was in, I realized this is indeed the future of knowledge work, so the professional in me took over. I realized that either I needed to get AI working for me, or I’d be working for AI.

That part of my AI skill set can be copied by anyone, regardless of experience level or age. Have a plan, dive in, and get good at it. That’s how professionals operate.

However, I’m also an experienced software engineer.

I started writing code at age 8. Almost 50 years ago.

That part of my skill set cannot be copied by just anybody, regardless of experience level or age. I can tell AI which direction to go and what the goals are because I’ve been there and done that. I have an idea of what this should look like at the end.

Also, I can tell when it’s going in the wrong direction. I can tell when it’s making stuff that might work, but won’t be robust, maintainable, or scale. I can tell when it’s hardcoding stuff that should be configurable. I can tell when it’s using resources it shouldn’t. I can tell when its making stuff that won’t ever support debugging (because, yes, there will still be bugs).

That’s 50 years of struggle, deep thinking, and writing the code myself. Difficult (not impossible) for junior developers to replicate.

So where does that leave our junior workforce?

I’m concerned, but I wonder if AI token cost might be the junior worker’s savior.

Here’s a cost comparison of using GPT-5.5 and Opus-4.7 for “heavy agentic coding” which I think is analogous to a junior developer’s role:

Junior Developer (loaded salary): $116k

GPT-5.5 (50M in / 10M out tokens): $145k

Opus-4.7 (50M in / 10M out tokens): $132k

This is for code development. Junior software engineers earn higher salaries than many other types of junior knowledge workers.

So if we compare across other industries, the AI costs stay about the same, but the junior worker costs move downward. And as I’ve mentioned before, AI costs will go up as the providers look IPO and generally start to cash in on their products.

Maybe good old-fashioned finances will solve the junior worker dilemma.